Preprints
https://doi.org/10.5194/amt-2022-199
https://doi.org/10.5194/amt-2022-199
 
21 Jul 2022
21 Jul 2022
Status: this preprint is currently under review for the journal AMT.

Dual-frequency spectral radar retrieval of snowfall microphysics: a physically constrained deep learning approach

Anne-Claire Billault-Roux1, Gionata Ghiggi1, Louis Jaffeux2, Audrey Martini3, Nicolas Viltard3, and Alexis Berne1 Anne-Claire Billault-Roux et al.
  • 1Environmental Remote Sensing Laboratory, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
  • 2Laboratoire de Météorologie Physique, Aubière, France
  • 3Laboratoire Atmosphère, Milieux et Observations Spatiales, IPSL, UVSQ Université Paris-Saclay, Sorbonne Université, CNRS, Guyancourt, France

Abstract. The use of meteorological radars to study snowfall microphysical properties and processes is well established, in particular through two techniques: the use of multi-frequency radar measurements and the analysis of radar Doppler spectra. We propose a novel approach to retrieve snowfall properties by combining both techniques, while relaxing some assumptions on e.g. beam matching and non-turbulent atmosphere.

The method relies on a two-step deep-learning framework inspired from data compression techniques: an encoder model maps a high-dimensional signal to a lower-dimensional “latent” space, while the decoder reconstructs the original signal from this latent space. Here, Doppler spectrograms at two frequencies constitute the high-dimensional input, while the latent features are constrained to represent the snowfall properties of interest. The decoder network is first trained to emulate Doppler spectra from a set of microphysical variables, using simulations from the radiative transfer model PAMTRA as training data. In a second step, the encoder network learns the inverse mapping, from real measured dual-frequency spectrograms to the microphysical latent space; doing so, it leverages the spatial consistency of the measurements to mitigate the problem's ill-posedness.

The method was implemented on X- and W-band data from the ICE GENESIS campaign that took place in the Swiss Jura in January 2021. An in-depth assessment of the retrieval’s accuracy was performed through comparisons with colocated aircraft in-situ measurements collected during 3 precipitation events. The agreement is overall good and opens up possibilities for acute characterization of snowfall microphysics on larger datasets. A discussion of the method's sensitivity and limitations is also conducted. 

The main contribution of this work is on the one hand the theoretical framework itself, which can be applied to other remote sensing retrieval applications and is thus possibly of interest to a broad audience across atmospheric sciences. On the other hand, the retrieved seven microphysical descriptors provide relevant insights into snowfall processes.

Anne-Claire Billault-Roux et al.

Status: final response (author comments only)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Anne-Claire Billault-Roux et al.

Anne-Claire Billault-Roux et al.

Viewed

Total article views: 382 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
281 95 6 382 3 4
  • HTML: 281
  • PDF: 95
  • XML: 6
  • Total: 382
  • BibTeX: 3
  • EndNote: 4
Views and downloads (calculated since 21 Jul 2022)
Cumulative views and downloads (calculated since 21 Jul 2022)

Viewed (geographical distribution)

Total article views: 400 (including HTML, PDF, and XML) Thereof 400 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 27 Nov 2022
Download
Short summary
Better understanding and modeling snowfall properties and processes is relevant to many fields, ranging from weather forecasting to aircraft safety. Remote sensing, and especially weather radars, can be used to gain insights into the microphysics of snowfall. In this work, we propose a new method to retrieve snowfall properties from radar measurements. It relies on an original deep learning framework, which is constrained with knowledge of the underlying physics i.e. electromagnetic scattering.